Ember Sync allows your Ember.js application to seamlessly work online and offline.
Alpha version: beware of the risks if you want to try it in production.
This README documents the state of master, probably there is features described here that do not reflect the status of previous/latest releases.
Ember Sync has a queue of operations. Every time you save, update or delete a record in your offline database (e.g LocalStorage, IndexedDB etc) as you usually do, a new operation is created in the queue.
The queue is processed in order and each operation is executed against the online store. If internet goes offline or your server is down, the queue stops being processed. No operations will be run concurrently and the order will always be respected.
While you have no internet connection, you can continue using your app and all data will be saved offline while new operations will continue being enqueued. When internet is restored, the queue will continue being processed and pending records will be created or updated against the online store.
- Ember Data is required (tested against 1.0.0-beta.8).
- You should generate a unique ID for your records (e.g UUID) in the client and backend to maintain consistency between offline and online data.
Whenever you search something, Ember Sync will automatically query concurrently both offline and online stores and merge the results into a single Ember array.
This means that findQuery
returns an Ember.A()
right away instead of a
promise. If 2 seconds later a response comes from the REST api, Ember Sync
will push new records to that array that was already being shown in the
template. This mimics how streams work.
In the process, Ember Sync will automatically update your offline database with the records that come from the online store.
For this example, we'll consider
IndexedDBAdapter
as offline
database and ActiveModelAdapter
as online one. You can use any databases you
want.
For Ember runtime, we'll use Ember CLI.
- Run
npm install --save ember-sync
in your project
-
Add
"ember-sync": "^0.1.0"
as a dependency in yourbower.json
file. -
Add the following to you
Brofile.js
and restart your server:app.import('vendor/ember-sync/dist/ember-sync.js', { 'ember-sync': [ 'default' ] });
-
Define an online store: to setup Ember Sync you have to define what is your online store. It can be done in many ways and it's not related to Ember Sync itself, but here's an example:
// app/initializers/online-store.js var CustomOnlineSerializer = DS.ActiveModelSerializer.extend(); var CustomOnlineAdapter = DS.ActiveModelAdapter.extend({ serializer: CustomOnlineSerializer.create(), namespace: '/api/v1' // your server namespace }); var OnlineStore = DS.Store.extend({ adapterFor: function(type) { return this.container.lookup('adapter:_custom_store'); }, serializerFor: function(type) { return this.container.lookup('serializer:_custom_store'); } }); export default { name: "onlineStore", initialize: function(container, application) { CustomOnlineSerializer.reopen({ container: container }); container.register('store:online', OnlineStore); container.register('adapter:_custom_store', CustomOnlineAdapter); container.register('serializer:_custom_store', CustomOnlineSerializer); application.inject('route', 'onlineStore', 'store:online'); application.inject('controller', 'onlineStore', 'store:online'); } };
If you don't need a custom serializer you don't need to define it as in previous example.
-
Define an offline store: just define an adapter and serializer as you would normally do. This offline store will be located at
this.store
by default.You may use IndexedDBAdapter, LocalStorageAdapter or others.
Important: if you're using an offline store that requires you to manually create tables/object stores, create
emberSyncQueueModel
. For example, LSAdapter will create it automatically for you, whereas in IndexedDB you will have to do it yourself. -
Define an Ember.js initializer: now we connect Ember Sync with your App.
If you are using the ember-addon version(installed with npm in am ember-cli > 0.0.36) project, you can skip this step.
// app/initializers/ember-sync.js import { default as EmberSync } from 'ember-sync'; export default { name: "ember-sync", initialize: function(container, application) { container.register("lib:emberSync", EmberSync); container.register('model:ember-sync-queue-model', EmberSync.create().get('queueModel')); application.inject('route', 'emberSync', "lib:emberSync"); application.inject('controller', 'emberSync', "lib:emberSync"); } };
For now, this is a manual step. Later, we'll automate this.
-
Wire Ember Sync with the stores: finally, tell Ember Sync what are your stores instances in your application route. If you're not using Ember CLI, this is your
ApplicationRoute
:// app/routes/application.js export default Ember.Route.extend({ init: function() { this._super(); this.emberSync.set('offlineStore', this.store); this.emberSync.set('onlineStore', this.onlineStore); this.emberSync.synchronizeOnline(); } });
We have to do it manually because we don't know the name of your online store.
Important: remember to call
this.emberSync.synchronizeOnline()
. This will activate the loop that will take care of the queue.
To save records, do the following:
var user, book;
user = this.emberSync.createRecord('user', { name: 'Robinson Crusoe' });
book = this.emberSync.createRecord('book', { name: 'The Life of Robinson Crusoe', });
user.get('books').pushObject(book);
user.emberSync.save().then(function(user) {
book.emberSync.save();
});
createRecord
and deleteRecord
, which are commonly called on this.store
,
are now called on this.emberSync
. From there on, just call .emberSync.save()
on the
models as you normally would.
In the example above, user
and book
could be used in your controllers just
fine. this.emberSync.createRecord
returns essentially the same
as this.store.createRecord
. For instance:
// app/routes/user.js
export default Ember.Route.extend({
model: function() {
return this.emberSync.createRecord('user', {name: 'Robinson'});
}
});
To find records, just do:
users = this.emberSync.findQuery('user', {name: 'Robinson Crusoe'});
This will automatically search for that user in both offline and online stores.
Remember that findQuery
doesn't return a promise, but Ember.A()
instead.
For instance, if a record is found offline in 50 miliseconds, it'll show up in your template
right away. If another record is then found in your RESTAdapter-based store 3
seconds later, the
array is automatically populated, simulating streams.
Instead of findQuery
, you can use find
as well, but it will return a
Promise:
user = this.emberSync.find('user', 1);
To delete a record, do the following:
this.emberSync.deleteRecord('user', record);
record.emberSync.save();
When doing findQuery
or find
, Ember Sync will automatically save the records
found online in the offline store.
Speaking about data consistency, it's very unlikely that you will ever have problems due to the the use of a queue in between offline and online envs.
Imagine your REST api takes 10 seconds to respond. You create a record, then updates it and, finally, deletes it. The online server hasn't even responded the creation request yet. Ember Sync is smart enough: it will enqueue creation, then update and then deletion in the online store, and it will run them in that specific order, even if the record doesn't exist offline anymore.
The reason for that is that when you do your update locally, you might have changed an associated model as a consequency. You want that side-effect to happen in the server too. The order of which you do operations matters. If the REST api goes down, Ember Sync will stop the queue and resume later in the same order.
Ember Sync will not, however, solve the problem of
computer 1 and computer 2 are offline, record 1 is altered in both
This sort of conflict should be dealt by your backend. Although we plan to build an API to help you figure that out in the client, we advise you to architecture your network of clients to avoid such conflicts in the first place.
The following are planned features and areas that need improvements.
- periodicaly and asynchronously download data from the online store and push the records into the offline store, so the user can have a enough data to go offline.
- more granular configuration on how conflicts should be handled.
- refactor code to smaller objects.
- encapsulate initializer so user doesn't have to write it explicitly.
First, install depdendencies with npm install
and bower install
. Then run
broccoli serve
.
Visit http://localhost:4200/tests
.
If you prefer, use npm run-script test-all
in your terminal. You need to
have PhantomJS installed.
To build a new version, just run npm run-script build
. The build will be
available in the dist/
directory.
Copyright (c) 2014 Alexandre de Oliveira MIT Style license. http://opensource.org/licenses/MIT